We are interested in finding the amino acids and structural features which determine water permeability in the aquaporin (AQP) family of proteins and in the larger family of membrane intrinsic proteins (MIPs). We have expressed the AQP1 protein in yeast secretory vesicles and have also reconstituted yeast-expressed AQP1 into lipid vesicles. We have shown that the protein acts as a water channel in both environments. We are in the process of expressing AQP2 and AQP3 in the same yeast system. We have expressed two closely related proteins (FPS and GlpF) from the MIP family which, although homologous in sequence, do not act as water channels. We would like to find specific amino acids or protein domains which differ among the water channels and non-water channels and which may be important for determining water permeability. To do this, we plan to analyze the sequences and secondary structures of all of the MIP and AQP proteins and look for regions which are conserved or variable. Sites which potentially determine water permeability can then be evaluated experimentally using our yeast expression system. Sequence alignments and secondary structure predictions are possible with the CGC software package which is available from the Pittsburgh Supercomputing Center.